On the Feasibility of Reference-Free Tracking via Motion Sensors in Public Transport
Smartphone motion sensors have been identified as a potential source of privacy leakage regarding user mobility in public transport networks. However, prior trajectory inference attacks often rely on strong reference data, including prior recordings on the target transport network and fixed phone placement, limiting the understanding of real-world attack fea sibility. In this work, we investigate the feasibility of reference free trajectory inference from smartphone motion sensors un der realistic sensing and attacker assumptions. We design an unsupervised inference pipeline in which the attacker relies only on publicly available transport information combined with structural features extracted from accelerometer and gyroscope signals in order to progressively reduce the set of possible trajectories and apply a plausibility-based ranking. We evaluate the proposed framework on 172 real-world motion sensor record ings collected across 21 transport lines in two German public transport networks. Our findings indicate that exact trajectory reconstruction without reference information remains challenging in practice. Nevertheless, coarse motion sensor observations still reveal meaningful mobility information and can substantially reduce trajectory uncertainty, particularly in structurally dis tinctive transport corridors.
